1. No category

BMC Pharmacology and
Toxicology
This Provisional PDF corresponds to the article as it appeared upon acceptance. Fully formatted
PDF and full text (HTML) versions will be made available soon.
Population pharmacokinetic and pharmacodynamic modeling of transformed
binary effect data of triflusal in healthy Korean male volunteers: a randomized,
open-label, multiple dose, crossover study
BMC Pharmacology and Toxicology 2014, 15:75
doi:10.1186/2050-6511-15-75
Sung Min Park (psm7271@naver.com)
Joomi Lee (joomibbc@hanmail.net)
Sook-Jin Seong (jy-nee@hanmail.net)
Jong Gwang Park (freesstyler@hanmail.net)
Mi-Ri Gwon (babymil1122@hanmail.net)
Mi-sun Lim (mslim@yu.ac.kr)
Hae Won Lee (haewonbbc@gmail.com)
Young-Ran Yoon (yry@knu.ac.kr)
Dong Heon Yang (ddhyang@knu.ac.kr)
Kwang-Il Kwon (kwon@cnu.ac.kr)
Seunghoon Han (waystolove@catholic.ac.kr)
ISSN
Article type
2050-6511
Research article
Submission date
21 May 2014
Acceptance date
12 December 2014
Publication date
23 December 2014
Article URL
http://www.biomedcentral.com/2050-6511/15/75
Like all articles in BMC journals, this peer-reviewed article can be downloaded, printed and
distributed freely for any purposes (see copyright notice below).
Articles in BMC journals are listed in PubMed and archived at PubMed Central.
For information about publishing your research in BMC journals or any BioMed Central journal, go to
http://www.biomedcentral.com/info/authors/
© 2014 Park et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain
Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Population pharmacokinetic and pharmacodynamic
modeling of transformed binary effect data of
triflusal in healthy Korean male volunteers: a
randomized, open-label, multiple dose, crossover
study
Sung Min Park1,2,3
Email: psm7271@naver.com
Joomi Lee1,2
Email: joomibbc@hanmail.net
Sook-Jin Seong1,2,3
Email: jy-nee@hanmail.net
Jong Gwang Park1
Email: freesstyler@hanmail.net
Mi-Ri Gwon1,2,3
Email: babymil1122@hanmail.net
Mi-sun Lim4
Email: mslim@yu.ac.kr
Hae Won Lee1
Email: haewonbbc@gmail.com
Young-Ran Yoon1,2,3
Email: yry@knu.ac.kr
Dong Heon Yang5
Email: ddhyang@knu.ac.kr
Kwang-Il Kwon6*
*
Corresponding author
Email: kwon@cnu.ac.kr
Seunghoon Han7,8*
*
Corresponding author
Email: waystolove@catholic.ac.kr
1
Clinical Trial Center, Kyungpook National University Hospital, Daegu, South
Korea
2
Department of Biomedical Science, Kyungpook National University Graduate
School, Daegu, South Korea
3
BK21 Plus KNU Bio-Medical Convergence Program for Creative Talent,
Kyungpook National University Graduate School, 680 Gukchaebosang-ro, Junggu, Daegu 700-842, South Korea
4
College of Pharmacy, Yeungnam University, Daegu, South Korea
5
Department of Internal Medicine, Division of Cardiology, Kyungpook National
University School of Medicine, Daegu, South Korea
6
College of Pharmacy, Chungnam National University, Daejeon, South Korea
7
Department of Pharmacology, College of Medicine, The Catholic University of
Korea, Seoul, South Korea
8
PIPET (Pharmacometrics Institute for Practical Education and Training), Seoul,
South Korea
Abstract
Background
Triflusal is a drug that inhibits platelet aggregation. In this study we investigated the doseexposure-response relationship of a triflusal formulation by population pharmacokinetic (PK)
and pharmacodynamic (PD) modeling of its main active metabolite, hydroxy-4(trifluoromethyl) benzoic acid (HTB).
Methods
This study was a randomized, open-label, multiple-dose, two-period, two-treatment,
comparative crossover design. All volunteers received a single oral loading dose of 900 mg
of triflusal on Day 1, followed by a dose of 600 mg/day from Day 2 to 9. Using data from 34
healthy volunteers, 476 HTB plasma concentration data points and 340 platelet aggregation
data points were used to construct PK and PD models respectively using NONMEM (version
6.2). As the PD endpoint was qualitative, we implemented binary analysis of ‘inhibition’ and
‘non-inhibition’ rather than using the actual value of the test. The final PK-PD model was
evaluated using a visual predictive check (VPC) and bootstrap.
Results
The time-concentration profile of HTB over the entire dosing period was described by a onecompartment model with a first-order formation rate constant for HTB. Weight was selected
as a covariate for clearance and volume of triflusal, respectively. The structure and the
population estimates for triflusal PK were as follows: oral clearance (CL/F) = 0.2 ·
(weight/71.65)0.845 L/h, oral volume of distribution (V/F) = 8.3 · (weight/71.65) L, and kf =
0.341 h−1. A sigmoid relationship between triflusal concentration and the probability of
significant inhibition with shape factor was chosen as the final PD model. No time delay
between concentration and response was identified. The final structure between predicted
concentration (Cpred,ijγ) and the probability of inhibition of platelet aggregation (IPA)
relationship was as follows: Probability of IPA = Cpred,ij19 / (84.9 µg/mL19 + Cpred,ij19). Thus,
we concluded this relationship is more like quantal concentration-response relationship. The
current dosing regimen was considered to be efficacious based on the EC50 estimate of 84.9
µg/mL obtained in this study.
Conclusions
A PK and binary probability PD model of triflusal was successfully developed for Korean
healthy volunteers. The model may be used to further prediction inhibition of platelet
aggregation by triflusal.
Trial registration
Clinical Research information Service (CRIS), KCT0001299 (Registered December 5, 2014)
Keywords
HTB, Transformed binary data, Platelet aggregation, Binary probability model, NONMEM,
Population pharmacokinetics and pharmacodynamics
Background
Triflusal (2-acetoxy-4-(trifluoromethyl) benzoic acid), which is chemically related to
salicylate [1] but is not a derivative of acetylsalicylic acid, is an antiplatelet drug [2,3] that
selectively inhibits arachidonic acid (AA) metabolism in platelets by irreversibly inhibiting
cyclooxygenase-1 (COX-1) and reduces thromboxane B2 (TXB2) production [4,5]. 2Hydroxy-4-(trifluoromethyl) benzoic acid (HTB), the main active metabolite of triflusal, is
formed by deacetylation upon passage through the liver [5]. Triflusal is less effective in
inhibiting COX-1 and reducing TXB2 than aspirin, but more effective than HTB [4]. HTB
increases the effects of triflusal on COX-1 inhibition compared with salicylic acid, the main
metabolite of aspirin, which competes with the prodrug for the active site on the
cyclooxygenase enzyme [5,6].
Several previous reports have described the pharmacokinetic (PK) characteristics of triflusal.
After oral administration, triflusal is absorbed rapidly within the small intestine, showing an
absolute bioavailability of 83–100% [7]. The plasma half-life (T1/2) is 0.5 ± 0.1 h for triflusal
and 34.4 ± 0.1 h for HTB [4]. More than 60% of the parent drug is eliminated by the kidney
[7] and the values for renal clearance for triflusal and HTB were found to be 0.8 ± 0.2 L/h
and 0.18 ± 0.04 L/h, respectively [5,8]. HTB reaches steady-state levels after 8–10 days of
treatment [5].
The aim of this study was to quantitatively determine the dose-exposure-response
relationship of triflusal in healthy Korean male subjects using a population analysis.
Additionally, we suggest a method of binary analysis that might be applied to platelet
aggregation data to identify pharmacodynamic (PD) markers.
Methods
Inclusion and exclusion criteria
Volunteers aged 20–55 years and within 20% of their ideal body weight [IBW (kg) = height
(cm) − 100) × 0.9] were enrolled in the study. Those who were not suitable on the basis of
physical examinations and routine laboratory tests (blood hematology, biochemistry,
prothrombin time, bleeding time, urinalysis) were excluded.
Study design and ethical considerations
The analysis was performed using data from a randomized, open-label, multiple-dose, twoperiod, two-treatment, comparative crossover study involving 38 healthy adult males. The
clinical trial was conducted at Kyungpook National University Hospital Clinical Trial Center
(KNUH CTC) to determine the bioequivalence and non-inferiority of two different triflusal
formulations, as reported previously [4].
The study was conducted in accordance with the guidelines of Good Clinical Practice and the
Declaration of Helsinki and its amendments at the KNUH CTC, and was approved by the
institutional review board at KNUH. Written informed consent was obtained from all
volunteers before their participation.
Study drugs and dosage regimen
Triflusal capsule (Disgren capsule, 300 mg), as the reference formulation, and triflusal EC
capsule (Disgren enteric-coated capsule, 300 mg), as the test formulation were manufactured
and provided by Myung-In Pharm. Co., Ltd (Seoul, Republic of Korea). The subjects
received the test or reference formulation in multiple doses, followed by a 13-day washout
period and subsequent administration of the alternative formulation. During each period, each
subject received a single 900 mg oral loading dose on Day 1, followed by a 600 mg/day
maintenance dose (given as two 300 mg capsules once daily) from Day 2 to 9.
Blood sampling procedures
For measurement of the plasma concentration of HTB (PK), whole blood samples were
obtained at time 0 (pre-dose), and 24, 48, 96, 144, 168, 192, 192.5, 193, 194, 196, 199, 202,
and 216 h after the first dose (from Day 1 to Day 10). For platelet aggregation measurements
(PD), sampling was performed at time 0 (pre-dose), and 24, 48, 96, 144, 168, 192, 196, 202,
and 216 h post-dose. The overall sampling schema is presented in Figure 1.
Figure 1 PK and PD study designs. Subjects (n = 34) eligible for this study were given a
900 mg loading dose on day 1, followed by a 600 mg/day maintenance dose on days 2, 3, 5,
7, 8, and 9. For PK assessment, blood samples were obtained at time 0, 24, 48, 96, 144, 168,
192, 192.5, 193, 194, 196, 199, 202 and 216 h. For PD assessment, blood samples were
obtained at time 0, 24, 48, 96, 144, 168, 192, 196, 202, and 216 h. PK, pharmacokinetic; PD,
pharmacodynamic; IPA, inhibition of platelet aggregation.
Blood (7 mL) for PK analysis was collected into tubes containing sodium heparin, and blood
(3 mL) for PD analysis was collected into tubes containing 0.109 mmol/L sodium citrate. An
aliquot (1 mL) of each plasma sample was placed in a microcentrifuge tube containing 0.1
mL 0.1 M HCl to prevent degradation of triflusal in the plasma. The remaining samples were
chilled promptly on crushed ice to maintain the distribution ratio between triflusal and HTB
[9].
HTB plasma concentration measurements
Concentrations of HTB, rather than triflusal, were measured in this study because the level of
triflusal is too low to be detected after 4 h when administered orally, as reported by Lee et al.
[4,10]. Blood samples were analyzed by high-performance liquid chromatography coupled
with tandem mass spectrometry (HPLC-MS/MS) [7]. Quantification was performed by
multiple reaction monitoring (MRM) transitions. A C18 column was used to separate the
components of the samples by chromatography. Linear calibration curves were analyzed in
the range of 1 to 300 µg/mL HTB (coefficient of correlation, r = 0.999). The lower limit of
quantification of HTB was 1 µg/mL. The intra-day and inter-day % CVs for HTB were each
less than 15%. Figure 2 shows the plasma concentrations of HTB with respect to time.
Figure 2 Raw data plot of time-concentration-response profile. (a) Profiles of observed
individual and median HTB plasma concentration over time. (b) Observed concentration and
platelet aggregation at the time of PD sampling.
Platelet aggregation assessment
The tubes containing sodium citrate were centrifuged (1,000 rpm, 10 min) and 500 µL of
platelet-rich plasma (PRP) was collected as the supernatant. The PRP was transferred to
microcentrifuge tubes, which were capped to prevent changes in pH upon exposure to air, and
the samples were kept on ice. To obtain 500 µL of platelet-poor plasma (PPP), the remaining
sample in the original tubes was centrifuged again (3,000 rpm, 10 min) [11]. Platelet
aggregation in PRP was assessed using a Chronolog optical aggregometer (Model 490-4D;
Chrono-log, Havertown, PA, USA). Cuvettes were lined up along the two rows of the
aggregometer, filled with 250 µL of PRP or PPP, and warmed for 5 min at 37°C [12]. To
measure the degree of platelet aggregation, 0.5 µL AA was added to the cuvettes as an
agonist, and light transmission aggregometry was observed for 5 min. The aggregometer was
calibrated with a cuvette containing PRP, equaling 0% light transmission, as the baseline and
with a second cuvette containing PPP, equaling 100% light transmission, as the reference
[13]. Based on analyses of the blank plasma samples, which were validated in three
individuals, the intra-day and inter-day precisions were calculated as (SD/mean) × 100 (%)
and ranged from 2.1 to 6.8% and from 2.1 to 7.6%, respectively [4].
Dataset
In the current study, PK and PD data from 34 healthy volunteers who received the reference
drug were used for population PK-PD analysis. In total, 476 HTB plasma concentration data
points and 340 platelet aggregation (%) data points were included. The latent variables used
as candidate covariates were age, height, and weight, and levels of albumin, creatinine, serum
aspartate transaminase (AST), and alanine transaminase (ALT). Creatinine clearance (CLCR),
calculated using the Cockcroft-Gault equation, was also included.
Population PK-PD model development
The population PK-PD analysis was conducted by non-linear mixed effects modeling using
NONMEM (ver. 6.2; Icon Development Solution, Ellicott City, MD, USA). The estimate and
the between-subject variability for PK-PD parameters were investigated. The first-order
conditional estimation (FOCE) method with interaction option was used whenever applicable
[14]. The between-subject variability of each parameter of the basic model was applied using
the exponential model:
Pij = θ j × exp (ηij )
where Pij is the value of the jth parameter in the ith individual, θj is the typical value of the jth
population parameter, and ηij is a random variable for the ith individual in the jth parameter. It
is assumed that ηij is normally distributed, with a mean of zero and a variance of ω2, and that
ηij ≈ N(0, ωij2). The residual variability, consisting of intra-individual variability,
experimental errors, process noise, and/or model mis-specifications, was modeled using
additive (εadd,ij), proportional (εpro,ij), and combined error structures. The combined additive
and proportional error model on the following equation was initially applied to explain the
gaps between the observed values and those predicted by individual PK parameters:
Cij = C pred ,ij × (1 + ε pro,ij ) + ε add ,ij
where Cij is the jth observed value in the ith individual, and εpro,ij and εadd,ij are the residual
intra-individual variability with a mean of zero and variances of σpro2 and σadd2, respectively.
A similar approach was made to the PD observations. The standard errors, which give
information on the precision of the parameter estimates, were explored using NONMEM’s
“$COVARIANCE” step.
Models were evaluated using diagnostic scatter plots, goodness-of-fit plots, and the log
likelihood ratio test (LRT). The results for LRT were considered statistically significant if the
objective function value (OFV) decreased by 3.84 (p = 0.05); this was used as a cut-off
criterion for model improvement.
For PK, various absorption functions and distributional (1-compartment vs. multicompartment) models were compared. When linking PK-PD relationship, two different
models – binary probability and inhibition of build-up – were tried sequentially to find the
best description on the PD data. The binary probability model uses a form of relationship
similar to the Michaelis-Menten equation, Cpred,ijγ / (EC50 γ + Cpred,ijγ) to explain instantaneous
PK-PD relationship. In this equation, Cpred,ij is the estimated concentration over time for
individuals, EC50 is the plasma concentration corresponding to the half-maximal response,
and γ is the shape factor. To implement this approach, the PD data (% platelet aggregation)
was transformed to a binary format based upon the finding that the distribution of PD
observation (the maximum platelet aggregation observed in each sample) was clearly
bimodal (Figure 2). The values above 74% were regarded as no inhibition of platelet
aggregation (IPA), and thus as DV = 0, whereas values below 74 were regarded as IPA (DV
= 1), according to a guideline presenting the normal ranges for the degree (%) of aggregation
in platelet-rich plasma [15]. To find a reliable initial estimate for this model, a logistic
regression procedure was performed in advance. For a binary output variable, Y, we can
model conditional probability P(Y = 1 | X = x) as a function of x [16]. Let π(x) be a linear
function of x, π(x) = P(Y = 1 | X = x). The model can be written as π(x) = exp(β0 + β1x) /
exp(β0 + β1x) + 1. Equivalently, by a logit transformation, the logistic regression model has a
linear relationship, logit(π(x)) = log(π(x) / 1 − π(x)) = β0 + β1x, where β0 is the intercept
coefficient and β1 is the slope coefficient, for which the sign is positive or negative if π(x) is a
strictly increasing or decreasing function of x, respectively [17]. Using the frequency of DV =
1 at each time point, the intercept (β0) and slope (β1) were estimated using SAS software (ver.
9.2; SAS Institute, Inc., Cary, NC, USA). The inhibition of build-up model was tried to
evaluate whether there is a time-delay between concentration and response in the form of
continuous variable. The differential equation for platelet aggregation at a certain HTB
concentration is as follows [18]:
dR / dt = kin × [1 − ( I max ·C ) / ( IC50 + C )] − kout × R
where dR/dt is the rate of change of the measured response (R) over time, kin is the turnover
input rate for production of the response, Imax is the fraction representing the maximal
capacity by which the drug can inhibit platelet aggregation (0 ≤ Imax ≤ 1), C is the HTB
concentration, IC50 is the median inhibition concentration, and kout is the turnover output rate
of the response.
Covariate analysis
The parameter-covariate relationship, which explains why PK and PD vary among
individuals, was explored using variables in the dataset. In the exploratory analysis, scatter
plots of one covariate versus another were used to assess correlations between covariates and
of parameters versus covariates. The numerical procedure used generalized additive modeling
(GAM), as implemented in the Xpose library (ver. 4.4.0, Department of Pharmaceutical
Biosciences of Uppsala University, Uppsala, Sweden) [19] of the ‘R’ software (ver. 2.15.3; R
Foundation for Statistical Computing, Vienna, Austria). In this procedure, a response random
variable, Y, is regressed on the individual covariates (Xj) according to the general equation:
E (Y | X 1 , X 2 , ···, X p ) = s0 + ∑ p j =1 s j ( X j )
where the s0 is an intercept, and s1(X1), · · ·, sp(Xp) are smooth functions that are estimated in
a nonparametric fashion. In the equation, candidate covariates with the greatest decrease in
the Akaike information criterion were selected as final covariates in the PK model when there
was a reduction of at least 3.84 compared with the previous model, and the relationship
between the random variable, η, for each parameter and the candidate covariates was
improved. Otherwise, they were dropped from the model.
Graphical analyses using goodness-of-fit plots – including observed-versus-populationpredicted, observed-versus-individual-predicted, and residual diagnostics – were also
conducted to diagnose the degree of model development.
Model evaluation
Bootstrap was recruited to check the robustness of the PK-PD model and parameter
estimates. The median value and 95% confidence interval for each parameter were obtained
from 1,000 bootstrap dataset. For the PK model, a visual predictive check (VPC) was
implemented to further evaluate the accuracy and predictive performance. A total of 1,000
replicate simulation was performed and the 50th (estimated population median), 5th, and 95th
percentiles of the created concentrations plotted against time were compared with the
overlaid observed concentrations. In addition, predicted concentration-response relationship
is plotted with the predicted probability of each concentration data observed at the time of PD
sampling.
Sample size determination
We considered the sample size for having an 80% power at a significance level of 0.05 for
bioequivalence trials. It was calculated under the condition of the equivalence limit of 22.3%,
the true mean difference of 10%, and the standard deviation of 24.2%. The calculation was
performed by a formula suggested from Chow et al. [20]. Volunteers were randomly assigned
to one of two sequences with simple randomization using SAS software (ver. 9.2; SAS
Institute, Inc., Cary, NC, USA).
Statistical analyses
The demographic data were presented using descriptive statistical analysis performed with
the standard SPSS package (version 12.0 for Windows, SPSS, College Station, TX, USA).
The mean, standard deviation, and range were obtained for each variable.
Results
Study population
Volunteers were recruited from August 2008 to September 2008. In total, thirty-eight
volunteers were enrolled in this study. Four subjects dropped out because of severe dental
pain, a medication administration error, a missing follow-up urinalysis, or platelet
aggregation 1% of baseline in the second period. Data from the remaining 34 subjects were
analyzed to construct the PK and PD models. According to the CONSORT flow diagram,
Figure 3 shows the subject disposition. The age, height, and weight (means ± SD) of the
subjects were 24.1 ± 1.7 years, 176.1 ± 4.9 cm, and 70.8 ± 9.0 kg, respectively (Table 1).
Figure 3 CONSORT flow diagram of subject disposition.
Table 1 Demographic information of the healthy volunteers
Group
Number of subjects (n = 34)
mean ± SD (Min – Max)
Age (years)
24.1 ± 1.7 (21 – 28)
Height (cm)
176.1 ± 4.9 (167.1 – 184.2)
Weight (kg)
70.8 ± 9.0 (53.3 – 89.7)
Population PK modeling
The time-concentration profile of HTB over the entire multiple-dosing period was best
described by a one-compartment model with first-order formation rate constant of HTB. This
model was implemented in the PREDPP library subroutine “ADVAN2 TRANS2” in
NONMEM. Residual variability was best described by the proportional error model, which
explained the difference between the predicted and observed values for individuals and
resulted in a significant decrease in the OFV. Weight was selected as the only covariate
candidate after GAM analysis which explains existing knowledge about the PK of the drug.
The effect of weight on CL/F and V/F was explained best with following each equation:
CL / F = θ1 × ( weight / 71.65 ) ,
θ4
V / F = θ 2 × ( weight / 71.65 )
where θ1 and θ2 were the estimated typical values of CL and V with a median weight (71.65)
and θ4 was the estimated influential factor for weight. Weight produced 30.782 (equivalent to
p < 0.001, 1 degree of freedom) decrease in the minimized OFV as well as the change in the
random BSV for CL/F from 18.5 % to 14.9 % and for V/F from 15.6 % to 9.5 % (as %CV).
Because only one demographic variable (weight) was selected as a meaningful covariate, the
process of backward covariate exclusion was omitted.
The goodness-of-fit plots for the final PK model are presented in Figure 4. Final estimates
from the PK model, explaining the mean value and the between-subject variability, are listed
in Table 2. The minimum concentration at steady state (Css,min) was estimated to be 103.5
µg/mL using the values of final population PK parameters; in addition, the estimated value of
the accumulation factor was 2.26.
Figure 4 Goodness-of-fit plots of the final population pharmacokinetic model. (a) OBS
versus PRED; (b) OBS versus IPRED; (c) |IWRES| versus IPRED; (d) CWRES versus timeafter the last dose. Circle: observation; solid black line: line of identity; gray line: loess
regression line. OBS, observation; PRED, population prediction; IPRED, individual
prediction; |IWRES|, absolute individual weighted residuals; CWRED, conditional weighted
residuals.
Table 2 Final Estimates of Population PK and PD Parameters
Parameter
Description (Units)
Fixed effect
Pharmacokinetic
CL/F = θ1 * (weight/71.65)θ4
θ1
θ4
Vd/F = θ2 * (weight/71.65)
θ2
TV of CL/F (L/h) for subject whose
weight = 71.65 kg
Exponent of weight proportional to CL/F
TV of Vd (L) for subject whose
weight = 71.65 kg
TV of kf (h−1)
Estimates from final model
Estimate
%RSEa
Bootstrap results
Median (95% CI)
0.200
2.7
0.1998 (0.1995 – 0.2002)
0.845
17.4
0.840 (0.830 – 0.850)
8.300
2.7
8.281 (8.267 – 8.295)
Shrinkage (%)
kf = θ 3
0.341
15.1
0.345 (0.341 – 0.348)
Pharmacodynamic
θ5
TV of EC50 (µg/mL)
84.9
4.0
85.19 (84.98 – 85.40)
θ6
TV of γ
19.2
22.4
20.70 (20.32 – 21.08)
BSV (%CV)
Pharmacokinetic
ω 12
BSV for CL/F
14.9
21.0
14.4 (14.3 – 14.5)
0.6
2
ω2
BSV for Vd
9.5
57.8
8.6 (8.4 – 8.8)
40.2
ω 32
BSV for kf
88.0
26.5
73.5 (72.8 – 74.1)
10.5
Pharmacodynamic
BSV for EC50
21.8
28.4
21.4 (21.2 – 21.5)
8.3
ω 42
2
ω5
BSV for γ
0 (Fixed)
NE
Residual error (σ2)b
Proportional error
0.098
6.5
0.0977 (0.0973 – 0.0981)
11.5
%RSE, % relative standard error; CI, confidence interval; CL/F, oral clearance; TV, typical value at population level; Vd/F, oral volume of distribution; kf, first-order
formation rate constant; EC50, HTB concentration at which the probability of IPA is 50%; γ, shape factor; BSV, between-subject variability; %CV, % coefficient of variation;
NE, not estimated.
a
%RSE is calculated from ω2.
b
Additive residual error was not selected.
Population PD modeling
The binary probability model was selected as the final PK-PD structure. The results of the
estimation from this PD model were as follows: γ = 19 and EC50 = 84.9 µg/mL. In addition to
these parameter estimates, information such as the observed and predicted number of times
that non-aggregation occurred at each time point is shown in Table 3. The between-subject
variability was only applied to EC50 and was well explained. The values estimated from the
PD model are summarized in Table 2.
Table 3 Observed and predicted IPA over time
Time Number of subjects
Number of observed IPA (%)
0h
34
1 (3%)
24 h
34
2 (6%)
48 h
34
11 (32%)
96 h
34
29 (85%)
144 h
34
25 (74%)
168 h
34
27 (79%)
192 h
34
30 (88%)
196 h
34
34 (100%)
202 h
34
34 (100%)
216 h
34
28 (82%)
IPA, inhibition of platelet aggregation.
Number of predicted IPA (%)
0 (0%)
2 (6%)
11 (32%)
25 (74%)
28 (82%)
30 (88%)
30 (88%)
34 (100%)
34 (100%)
30 (88%)
Model evaluation
The bootstrap results showed acceptable robustness of the model and are presented together
with the final estimates in Table 2. The VPC result (Figure 5) showed that the prediction of
the simulated data well-matched the observed concentration-time profiles. This graph,
representing a visual internal validation of the model, showed that most of the observed data
points were overlaid between the 5th and 95th percentiles.
Figure 5 Results of model evaluation. (a) Visual predictive check result. The 5th and 95th
percentiles (dashed gray line) and the median value (solid black line) from the simulated data
are plotted against the observed concentration data (circle) according to time. (b) Predicted
HTB concentration-probability of IPA curve; Circle: expected IPA for each HTB
concentration measured at the time of PD observation; IPA, inhibition of platelet aggregation.
Discussion
The PK and PD characteristics of triflusal were investigated using a non-linear mixed effects
analysis. The objective of this study was to develop a population PK and PD model of the
triflusal formulation, an established uncoated capsule, in healthy Korean male subjects.
Plasma concentrations of the active metabolite, HTB, and the degree of platelet aggregation
inhibition were measured to demonstrate the PK and PD properties. Reasonable estimates for
parameters of the PK model were obtained, reflecting the potential for accurately predicting
the information sought. The plots of the observed concentrations versus population-predicted
or individual-predicted concentrations showed that the PK model was well structured, and the
plot of time versus conditional weighted residuals suggested that the errors had homogeneous
variances.
Our final HTB PK structure includes first-order formation and 1-compartment disposition.
This is plausible judging from the immediate increase in HTB concentration on the day of
full-PK study after dosing time and the rate of concentration increase which is the fastest at
immediate post-dose and decreases thereafter (Figure 2). This finding is consistent to the
model used by Valle et al. [3]. On the contrary, Yun et al. [21] suggested a 2-compartment
model for HTB disposition. But we considered this model would behavior like a 1compartment model because the distributional rate constant between compartments were
much (approximately 60 times) larger than the elimination rate constant and the equilibrium
between compartments will be obtained within 1 h after dose. Judging from the time of
maximum plasma concentration attainment (postdose 4–5 h), this distributional characteristic
might be superimposed by the absorption as well as the elimination.
Covariate model building was established by the relationship between the subjects’ body
features and parameters in the PK model. Weight accounted for approximately 19 % and 39
% of between-subject variability for CL/F and V/F, respectively, judging from the decrease of
ω2 for corresponding parameters. Particularly, V/F was directly proportional to weight
(exponent = 1) and the fact suggested that weight may be used as a scaling factor for triflusal
dose. However, definite conclusions could not be made for following reason; 1) this study did
not consider the PK of triflusal; thus, weight might be influential to the disposition of triflusal
including metabolic fraction to HTB (which was only explained with F in this study), 2) the
clinical significance of this finding needs further investigations together with its
pharmacodynamics variability, therapeutic index and clinical effectiveness. In addition, even
though HTB is mainly (>60%) excreted in urine, CLCR was not selected as a meaningful
covariate for HTB CL/F. We consider that the effect of weight superimposed that of CLCR. It
was expected that studies involving triflusal and HTB PK in patients with various renal
function will be essential to elucidate the exact influence of renal function.
Based on the value of EC50 (=84.9 µg/mL) estimated in this study, the current dosing regimen
was considered to be efficacious judging from the Css,min value (=103.5 µg/mL) and the
concentration-response relationship which showed quantal characteristics (γ = 19). The
saturated value of Emax as 1 seemed reasonable, because there was no significant model
improvement when the parameter was allowed to be estimated.
Another method, the binary probability model to construct the PD model using transformed
binary data, was introduced to show characteristics of the PD data without much information
loss. This model resulted in successful explanation of binary data characteristics by the
estimates, even though a turnover model was not constructed because of the modified binary
data characteristics. To date, this is the first report evaluating quantitative PD data
transformed into qualitative binary data possessing one of two values. We recommend this
new approach of using probability to estimate parameters that show predictive ability for the
presence of platelet aggregation. This model also provided a specific method for potential use
with binary data in the case of high intra-individual variability for PD.
This study has some limitations. First, only HTB concentrations were used in our population
PK and PD model without measurement of the parent drug. Although triflusal is a prodrug
which has no pharmacologic effect, if a study involving the concentration measurement of
both triflusal and HTB is conducted, a better description for the HTB formation will be
obtained. This approach may enable the identification of influential factors for HTB
formation process including first-pass metabolism and hepatic clearance of triflusal. Second,
despite the possible loss of information from the data transformation, a PD model using
probability (binary data model) was developed to evaluate the extreme values in our PD data,
as seen in the scatter diagram. To overcome this problem, more accurate and precise
quantitative analytic method to measure the degree of platelet aggregation should be required.
Using the VPC, which is one method of model evaluation, prediction is possible when given
not only different dosage- and time-concentrations but also the probability of IPA.
Accordingly, optimal dosage regimens could be determined by model-fitted parameter
estimates without additional clinical trials. The model may be clinically useful for other
similar studies. Finally, this study was conducted in healthy volunteers, rather than in
patients. Thus, it is also necessary to construct a model estimating reasonable parameters
from patients in a more realistic setting.
Conclusions
A PK and binary probability PD model of triflusal was successfully developed for Korean
healthy volunteers. The model may be used to further predict inhibition of platelet
aggregation by triflusal.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
SMP constructed and developed the PK and PD models and drafted the manuscript. JL, SJS,
JGP, MRG, MSL, HWL, DHY, and YRY helped prepare datasets, develop the PK and PD
models, and interpret the results of the estimated parameters. KIK and SH designed and
supervised the clinical study, developed the PK and PD models, and edited the manuscript.
All authors read and approved the final manuscript.
Acknowledgements
This research was supported by grants from the National Project for Personalized Genomic
Medicine, Ministry of Health & Welfare, Republic of Korea (A111218-PG02); the Bio &
Medical Technology Development Program of the National Research Foundation (NRF)
funded by the Ministry of Science, ICT & Future Planning, Republic of Korea (NRF2013M3A9B6046416); the Pfizer Modeling & Simulation Education Center in Korea
(PMECK); and the Korea Health 21 R&D Project, Ministry of Health & Welfare, Republic of
Korea (A070001).
References
1. Matias-Guiu J, Ferro JM, Alvarez-Sabin J, Torres F, Jiménez MD, Lago A, Melo T:
Comparison of triflusal and aspirin for prevention of vascular events in patients after
cerebral infarction: the TACIP Study: a randomized, double-blind, multicenter trial.
Stroke 2003, 34:840–848.
2. McNeely W, Goa KL: Triflusal. Drugs 1998, 7:823–833.
3. Valle M, Barbanoj MJ, Donner A, Izquierdo I, Herranz U, Klein N, Eichler HG, Müller M,
Brunner M: Access of HTB, main metabolite of triflusal, to cerebrospinal fluid in healthy
volunteers. Eur J Clin Pharmacol 2005, 61:103–111.
4. Lee HW, Lim MS, Seong SJ, Lee J, Park J, Seo JJ, Yun HY, Baek IH, Kwon KI, Yoon
YR: A phase I study to characterize the multiple-dose pharmacokinetics,
pharmacodynamics and safety of new enteric-coated triflusal formulations in healthy
male volunteers. Expert Opin Drug Metab Toxicol 2011, 7:1471–1479.
5. Anninos H, Andrikopoulos G, Pastromas S, Sakellariou D, Theodorakis G, Vardas P:
Triflusal: an old drug in modern antiplatelet therapy. Review of its action, use, safety
and effectiveness. Hellenic J Cardiol 2009, 50:199–207.
6. Rao GH, Reddy KR, White JG: Effect of acetaminophen and salicylate on aspirininduced inhibition of human platelet cyclo-oxygenase. Prostaglandins Leukot Med 1982,
9:109–115.
7. Murdoch D, Plosker GL: Triflusal: a review of its use in cerebral infarction and
myocardial infarction, and as thromboprophylaxis in atrial fibrillation. Drugs 2006,
66:671–692.
8. Gonzalez-Correa JA, De La Cruz JP: Triflusal: an antiplatelet drug with a
neuroprotective effect? Cardiovasc Drug Rev 2006, 24:11–24.
9. Cho HY, Jeong TJ, Lee YB: Simultaneous determination of triflusal and its major
active metabolite, 2-hydroxy-4-trifluoromethyl benzoic acid, in rat and human plasma
by high-performance liquid chromatography. J Chromatogr B Analyt Technol Biomed
Life Sci 2003, 798:257–264.
10. Ramis J, Mis R, Forn J, Torrent J, Gorina E, Jane F: Pharmacokinetics of triflusal and
its main metabolite HTB in healthy subjects following a single oral dose. Eur J Drug
Metab Pharmacokinet 1991, 16:269–273.
11. Dong HP, Wu HM, Chen SJ, Chen CY: The effect of butanolides from Cinnamomum
tenuifolium on platelet aggregation. Molecules 2013, 18:11836–11841.
12. Harrison P, Mackie I, Mumford A, Briggs C, Liesner R, Winter M, Machin S: Guidelines
for the laboratory investigation of heritable disorders of platelet function. Br J Haematol
2011, 155:30–44.
13. Zhou L, Schmaier AH: Platelet aggregation testing in platelet-rich plasma:
description of procedures with the aim to develop standards in the field. Am J Clin
Pathol 2005, 123:172–183.
14. Beal S, Sheiner L: NONMEM User’s Guide Part I. San Francisco: University of
California at San Francisco; 1992.
15. Instruction manual for the chrono-log model 490 optical aggregometers. Havertown, PA,
USA: Chrono-log.
16. Agresti A: Logistic Regression. In Categorical Data Analysis. 2nd edition. New York:
Wiley; 2002:166.
17. Casella G, Berger RL: Regression Models. In Statistical Inference. 2nd edition.
California: Wadsworth; 2001:592.
18. Gabrielsson J, Weiner D: Pharmacodynamic Concepts. In Pharmacokinetic and
Pharmacodynamic Data Analysis: Concepts and Applications. 4th edition. Stockholm:
Swedish Pharmaceutical Press; 2007:265.
19. Jonsson EN, Karlsson MO: Xpose–an S-PLUS based population
pharmacokinetic/pharmacodynamic model building aid for NONMEM. Comput
Methods Programs Biomed 1999, 58:51–64.
20. Chow SC, Wang H: On sample size calculation in bioequivalence trials. J
Pharmacokinet Pharmacodyn 2001, 28:155–169.
21. Yun HY, Kang W, Lee BY, Park S, Yoon YR, Yeul Ma J, Kwon KI: Semi-Mechanistic
Modelling and Simulation of Inhibition of Platelet Aggregation by Antiplatelet Agents.
Basic Clin Pharmacol Toxicol 2014. in press.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5